[4.0.3 backport] CBG-5142: Only fetch backup revision for fromRev for delta sync (#8048)

Author: gregns1

db/attachment_test.go

@@ -134,7 +134,7 @@ func TestGetBackupRevisionWhenCurrentRevisionHasAttachments(t *testing.T) {
 	// can remove in CBG-4542
 	db.FlushRevisionCacheForTest()
 
-	docRev, err := collection.GetRev(ctx, "doc1", doc1.HLV.GetCurrentVersionString(), false, nil)
+	docRev, err := collection.revisionCache.GetWithCV(ctx, "doc1", doc1.HLV.ExtractCurrentVersionFromHLV(), false, true)
 	require.NoError(t, err)
 
 	// assert version is fetched and attachments is empty

db/crud.go

@@ -357,7 +357,7 @@ func (db *DatabaseCollectionWithUser) getRev(ctx context.Context, docid, revOrCV
 			revision, getErr = db.revisionCache.GetWithRev(ctx, docid, *revID, RevCacheOmitDelta)
 		} else {
 			cv = &currentVersion
-			revision, getErr = db.revisionCache.GetWithCV(ctx, docid, cv, RevCacheOmitDelta)
+			revision, getErr = db.revisionCache.GetWithCV(ctx, docid, cv, RevCacheOmitDelta, false)
 		}
 	} else {
 		// No rev given, so load active revision
@@ -432,7 +432,7 @@ func (db *DatabaseCollectionWithUser) documentRevisionForRequest(ctx context.Con
 
 func (db *DatabaseCollectionWithUser) GetCV(ctx context.Context, docid string, cv *Version, revTreeHistory bool) (revision DocumentRevision, err error) {
 	if cv != nil {
-		revision, err = db.revisionCache.GetWithCV(ctx, docid, cv, RevCacheOmitDelta)
+		revision, err = db.revisionCache.GetWithCV(ctx, docid, cv, RevCacheOmitDelta, false)
 	} else {
 		revision, err = db.revisionCache.GetActive(ctx, docid)
 	}
@@ -462,7 +462,8 @@ func (db *DatabaseCollectionWithUser) GetDelta(ctx context.Context, docID, fromR
 		if err != nil {
 			return nil, nil, err
 		}
-		initialFromRevision, err = db.revisionCache.GetWithCV(ctx, docID, &fromRevVrs, RevCacheIncludeDelta)
+		// It is possible delta source will not be resident in the cache and we may want to lookup to the bucket for a backup revision
+		initialFromRevision, err = db.revisionCache.GetWithCV(ctx, docID, &fromRevVrs, RevCacheIncludeDelta, true)
 		if err != nil {
 			return nil, nil, err
 		}
@@ -503,7 +504,7 @@ func (db *DatabaseCollectionWithUser) GetDelta(ctx context.Context, docID, fromR
 		// fromRevisionForDiff is a version of the fromRevision that is guarded by the delta lock that we will use to generate the delta (or check again for a newly cached delta)
 		var fromRevisionForDiff DocumentRevision
 		if fromRevIsCV {
-			fromRevisionForDiff, err = db.revisionCache.GetWithCV(ctx, docID, &fromRevVrs, RevCacheIncludeDelta)
+			fromRevisionForDiff, err = db.revisionCache.GetWithCV(ctx, docID, &fromRevVrs, RevCacheIncludeDelta, true)
 			if err != nil {
 				return nil, nil, err
 			}
@@ -537,7 +538,7 @@ func (db *DatabaseCollectionWithUser) GetDelta(ctx context.Context, docID, fromR
 			if err != nil {
 				return nil, nil, err
 			}
-			toRevision, err = db.revisionCache.GetWithCV(ctx, docID, &cv, RevCacheIncludeDelta)
+			toRevision, err = db.revisionCache.GetWithCV(ctx, docID, &cv, RevCacheIncludeDelta, false)
 			if err != nil {
 				return nil, nil, err
 			}

db/database_test.go

@@ -1090,14 +1090,14 @@ func TestFetchCurrentRevAfterFetchBackupRevByCV(t *testing.T) {
 		"k1":    "v2",
 		BodyRev: rev1ID,
 	}
-	rev2ID, _, err := collection.Put(ctx, "doc1", rev2Body)
+	rev2ID, doc2, err := collection.Put(ctx, "doc1", rev2Body)
 	require.NoError(t, err, "Error creating doc")
 
 	// Flush the revision cache, this can be removed pending CBG-4542
 	db.FlushRevisionCacheForTest()
 
 	// fetch backup rev by cv and ensure we have no revID populated (no way to get revID from backup rev in CV)
-	docRev, err := collection.GetRev(ctx, "doc1", doc.CV(), true, nil)
+	docRev, err := collection.revisionCache.GetWithCV(ctx, "doc1", doc.HLV.ExtractCurrentVersionFromHLV(), false, true)
 	require.NoError(t, err, "Error fetching backup revision CV")
 	assert.Equal(t, "", docRev.RevID)
 	assert.Equal(t, `{"k1":"v1"}`, string(docRev.BodyBytes))
@@ -1107,6 +1107,7 @@ func TestFetchCurrentRevAfterFetchBackupRevByCV(t *testing.T) {
 	require.NoError(t, err, "error fetching current revision")
 	assert.Equal(t, rev2ID, docRev.RevID)
 	assert.Equal(t, `{"k1":"v2"}`, string(docRev.BodyBytes))
+	assert.Equal(t, doc2.HLV.GetCurrentVersionString(), docRev.CV.String())
 }
 
 func TestFetchCurrentRevAfterFetchBackupRevByRevID(t *testing.T) {

db/hybrid_logical_vector_test.go

@@ -335,7 +335,7 @@ func TestHLVImport(t *testing.T) {
 			name: "HLV write from without mou",
 			preFunc: func(t *testing.T, collection *DatabaseCollectionWithUser, docID string) {
 				hlvHelper := NewHLVAgent(t, collection.dataStore, otherSource, "_vv")
-				_ = hlvHelper.InsertWithHLV(ctx, docID)
+				_ = hlvHelper.InsertWithHLV(ctx, docID, nil)
 			},
 			expectedMou: func(output *outputData) *MetadataOnlyUpdate {
 				return &MetadataOnlyUpdate{
@@ -356,7 +356,7 @@ func TestHLVImport(t *testing.T) {
 			name: "XDCR stamped with _mou",
 			preFunc: func(t *testing.T, collection *DatabaseCollectionWithUser, docID string) {
 				hlvHelper := NewHLVAgent(t, collection.dataStore, otherSource, "_vv")
-				cas := hlvHelper.InsertWithHLV(ctx, docID)
+				cas := hlvHelper.InsertWithHLV(ctx, docID, nil)
 
 				_, xattrs, _, err := collection.dataStore.GetWithXattrs(ctx, docID, []string{base.VirtualXattrRevSeqNo})
 				require.NoError(t, err)
@@ -387,7 +387,7 @@ func TestHLVImport(t *testing.T) {
 			name: "invalid _mou, but valid hlv",
 			preFunc: func(t *testing.T, collection *DatabaseCollectionWithUser, docID string) {
 				hlvHelper := NewHLVAgent(t, collection.dataStore, otherSource, "_vv")
-				cas := hlvHelper.InsertWithHLV(ctx, docID)
+				cas := hlvHelper.InsertWithHLV(ctx, docID, nil)
 
 				_, xattrs, _, err := collection.dataStore.GetWithXattrs(ctx, docID, []string{base.VirtualXattrRevSeqNo})
 				require.NoError(t, err)

db/revision_cache_bypass.go

@@ -58,7 +58,7 @@ func (rc *BypassRevisionCache) GetWithRev(ctx context.Context, docID, revID stri
 }
 
 // GetWithCV fetches the Current Version for the given docID and CV immediately from the bucket.
-func (rc *BypassRevisionCache) GetWithCV(ctx context.Context, docID string, cv *Version, collectionID uint32, includeDelta bool) (docRev DocumentRevision, err error) {
+func (rc *BypassRevisionCache) GetWithCV(ctx context.Context, docID string, cv *Version, collectionID uint32, includeDelta bool, loadBackup bool) (docRev DocumentRevision, err error) {
 
 	docRev = DocumentRevision{
 		CV:                     cv,
@@ -71,12 +71,11 @@ func (rc *BypassRevisionCache) GetWithCV(ctx context.Context, docID string, cv *
 	}
 
 	var hlv *HybridLogicalVector
-	docRev.BodyBytes, docRev.History, docRev.Channels, docRev.Removed, docRev.Attachments, docRev.Deleted, docRev.Expiry, docRev.RevID, hlv, err = revCacheLoaderForDocumentCV(ctx, rc.backingStores[collectionID], doc, *cv)
+	docRev.BodyBytes, docRev.History, docRev.Channels, docRev.Removed, docRev.Attachments, docRev.Deleted, docRev.Expiry, docRev.RevID, hlv, err = revCacheLoaderForDocumentCV(ctx, rc.backingStores[collectionID], doc, *cv, loadBackup)
 	if err != nil {
 		return DocumentRevision{}, err
 	}
 	if hlv != nil {
-		docRev.CV = hlv.ExtractCurrentVersionFromHLV()
 		docRev.HlvHistory = hlv.ToHistoryForHLV()
 	}
 

db/revision_cache_interface.go

@@ -34,9 +34,9 @@ type RevisionCache interface {
 	GetWithRev(ctx context.Context, docID, revID string, collectionID uint32, includeDelta bool) (DocumentRevision, error)
 
 	// GetWithCV returns the given revision by CV, and stores if not already cached.
-	// When includeBody=true, the returned DocumentRevision will include a mutable shallow copy of the marshaled body.
 	// When includeDelta=true, the returned DocumentRevision will include delta - requires additional locking during retrieval.
-	GetWithCV(ctx context.Context, docID string, cv *Version, collectionID uint32, includeDelta bool) (DocumentRevision, error)
+	// When loadBackup=true, will load from backup revisions if requested version is not active document
+	GetWithCV(ctx context.Context, docID string, cv *Version, collectionID uint32, includeDelta bool, loadBackup bool) (DocumentRevision, error)
 
 	// GetActive returns the current revision for the given doc ID, and stores if not already cached.
 	GetActive(ctx context.Context, docID string, collectionID uint32) (docRev DocumentRevision, err error)
@@ -127,7 +127,7 @@ func DefaultRevisionCacheOptions() *RevisionCacheOptions {
 type RevisionCacheBackingStore interface {
 	GetDocument(ctx context.Context, docid string, unmarshalLevel DocumentUnmarshalLevel) (doc *Document, err error)
 	getRevision(ctx context.Context, doc *Document, revid string) ([]byte, AttachmentsMeta, base.Set, error)
-	getCurrentVersion(ctx context.Context, doc *Document, cv Version) ([]byte, AttachmentsMeta, base.Set, bool, error)
+	getCurrentVersion(ctx context.Context, doc *Document, cv Version, loadBackup bool) ([]byte, AttachmentsMeta, base.Set, bool, error)
 }
 
 // collectionRevisionCache is a view of a revision cache for a collection.
@@ -150,8 +150,8 @@ func (c *collectionRevisionCache) GetWithRev(ctx context.Context, docID, revID s
 }
 
 // Get is for per collection access to Get method
-func (c *collectionRevisionCache) GetWithCV(ctx context.Context, docID string, cv *Version, includeDelta bool) (DocumentRevision, error) {
-	return (*c.revCache).GetWithCV(ctx, docID, cv, c.collectionID, includeDelta)
+func (c *collectionRevisionCache) GetWithCV(ctx context.Context, docID string, cv *Version, includeDelta bool, loadBackup bool) (DocumentRevision, error) {
+	return (*c.revCache).GetWithCV(ctx, docID, cv, c.collectionID, includeDelta, loadBackup)
 }
 
 // GetActive is for per collection access to GetActive method
@@ -421,7 +421,7 @@ func revCacheLoader(ctx context.Context, backingStore RevisionCacheBackingStore,
 
 // revCacheLoaderForCv will load a document from the bucket using the CV, compare the fetched doc and the CV specified in the function,
 // and will still return revid for purpose of populating the Rev ID lookup map on the cache
-func revCacheLoaderForCv(ctx context.Context, backingStore RevisionCacheBackingStore, id IDandCV) (bodyBytes []byte, history Revisions, channels base.Set, removed bool, attachments AttachmentsMeta, deleted bool, expiry *time.Time, revid string, hlv *HybridLogicalVector, err error) {
+func revCacheLoaderForCv(ctx context.Context, backingStore RevisionCacheBackingStore, id IDandCV, loadBackup bool) (bodyBytes []byte, history Revisions, channels base.Set, removed bool, attachments AttachmentsMeta, deleted bool, expiry *time.Time, revid string, hlv *HybridLogicalVector, err error) {
 	cv := Version{
 		Value:    id.Version,
 		SourceID: id.Source,
@@ -431,7 +431,7 @@ func revCacheLoaderForCv(ctx context.Context, backingStore RevisionCacheBackingS
 		return bodyBytes, history, channels, removed, attachments, deleted, expiry, revid, hlv, err
 	}
 
-	return revCacheLoaderForDocumentCV(ctx, backingStore, doc, cv)
+	return revCacheLoaderForDocumentCV(ctx, backingStore, doc, cv, loadBackup)
 }
 
 // Common revCacheLoader functionality used either during a cache miss (from revCacheLoader), or directly when retrieving current rev from cache
@@ -471,8 +471,8 @@ func revCacheLoaderForDocument(ctx context.Context, backingStore RevisionCacheBa
 
 // revCacheLoaderForDocumentCV used either during cache miss (from revCacheLoaderForCv), or used directly when getting current active CV from cache
 // nolint:staticcheck
-func revCacheLoaderForDocumentCV(ctx context.Context, backingStore RevisionCacheBackingStore, doc *Document, cv Version) (bodyBytes []byte, history Revisions, channels base.Set, removed bool, attachments AttachmentsMeta, deleted bool, expiry *time.Time, revid string, hlv *HybridLogicalVector, err error) {
-	if bodyBytes, attachments, channels, deleted, err = backingStore.getCurrentVersion(ctx, doc, cv); err != nil {
+func revCacheLoaderForDocumentCV(ctx context.Context, backingStore RevisionCacheBackingStore, doc *Document, cv Version, loadBackup bool) (bodyBytes []byte, history Revisions, channels base.Set, removed bool, attachments AttachmentsMeta, deleted bool, expiry *time.Time, revid string, hlv *HybridLogicalVector, err error) {
+	if bodyBytes, attachments, channels, deleted, err = backingStore.getCurrentVersion(ctx, doc, cv, loadBackup); err != nil {
 		return nil, nil, nil, false, nil, false, nil, "", nil, err
 	}
 
@@ -481,9 +481,9 @@ func revCacheLoaderForDocumentCV(ctx context.Context, backingStore RevisionCache
 	if doc.HLV.ExtractCurrentVersionFromHLV().Equal(cv) {
 		revid = doc.GetRevTreeID()
 		deleted = doc.Deleted
+		hlv = doc.HLV
 	}
 
-	hlv = doc.HLV
 	validatedHistory, getHistoryErr := doc.History.getHistory(revid)
 	if getHistoryErr != nil {
 		return bodyBytes, history, channels, removed, attachments, deleted, doc.Expiry, revid, hlv, err
@@ -493,8 +493,12 @@ func revCacheLoaderForDocumentCV(ctx context.Context, backingStore RevisionCache
 	return bodyBytes, history, channels, removed, attachments, deleted, doc.Expiry, revid, hlv, err
 }
 
-func (c *DatabaseCollection) getCurrentVersion(ctx context.Context, doc *Document, cv Version) (bodyBytes []byte, attachments AttachmentsMeta, channels base.Set, deleted bool, err error) {
+func (c *DatabaseCollection) getCurrentVersion(ctx context.Context, doc *Document, cv Version, loadBackup bool) (bodyBytes []byte, attachments AttachmentsMeta, channels base.Set, deleted bool, err error) {
 	if err = doc.HasCurrentVersion(ctx, cv); err != nil {
+		if !loadBackup {
+			// do not attempt to fetch backup revision by CV unless specified
+			return nil, nil, nil, false, ErrMissing
+		}
 		bodyBytes, channels, deleted, err = c.getOldRevisionJSON(ctx, doc.ID, base.Crc32cHashString([]byte(cv.String())))
 		if err != nil || bodyBytes == nil {
 			return nil, nil, nil, false, err